I mean---what if (1) calories in affects calories out, and (2) calories out affects calories in: eating more calories causes expenditure of more calories; and expending more calories causes eating of more calories.
What besides a complex system can possibly account for it? One answer is that the "system" is not exactly the same for the 3000 cal scenario and the 500 cal scenario.
It depends on what XYZ are. If to the body, X > Y > Z in importance, then the body that requires 3000 cals to do X+Y+Z but now only gets 500 cals, may only do X. But if the body continues to do XYZ on 500 cals, it may make several other changes, such as eliminating ABCDEFGH and cutting XYZ down to a minimum. These changes create a totally different system than the original, in other words, a different physiological environment on the latter (500 cals) than the former (3000 cals).
If Y or Z is exercise (voluntary large muscle movement), consider how efficient our large muscles are in terms of energy expenditure.
Cites please. And be skeptical of the skeptics (it's fun).
You didn't read the whole study. They differentiated between lost body water, lost body weight, and lost LBM, iirc. (All together these allow deductions about lost body fat.)
I hope you enjoy it. And be skeptical of what you read (because it's fun).